The present invention relates to event relationship analysis in fault management.
Data center and network management disciplines have focused on fault and root cause analysis processes, tools and best practices. When events occur in a data center, a notification is sent to an event manager. At the event manager, the event may be de-duplicated, correlated, and enriched. It may be handled via a policy enforced by a rules engine. It may be used to automatically create a ticket for a help desk. To reduce operation cost, it is known to correlate commonly co-occurring alerts together so as to allow an operator to only work on one problem or one ticket open for a single problem.
Event manager products may implement concepts for identifying recurrent patterns in an event stream so that when operators are presented with a set of incoming events the list may be compacted using previously-observed relationships. Such identification of event patterns may not be trivial and typically provides limited information. By way of example, identified patterns may be resource specific (e.g. ‘resource bound’) and/or require an event to occur multiple times in order for a pattern to be identified.